Often enterprise processes rely on human activities and cannot be fully automated. This results in unpredictable behavior in the course of process execution.
Typically, when anomalies are detected during process execution, alerts are generated. The process owners react to these alerts by reengineering or redesigning the process. This approach, however, is costly and time consuming. Besides, reactive approaches may not successfully change the course of process execution and hence may not stop anomalies.